[whatmodel.sh](../../README.md) / DeepSeek R1 Distill Qwen 32B # DeepSeek R1 Distill Qwen 32B DeepSeek · released 2025-01 · 32.8B params · mit **score 55.7** ±0.8 · #265 of 732 · from 28 benchmark results across 11 sources Score within the [LMArena (text)](https://whatmodel.sh/?set=lmarena-text&score=set) comparison set only. ## What it's good at ● marks DeepSeek R1 Distill Qwen 32B among every model on that task or test, worst to best. 
task    vs the others         rank  counts  tests
coding  ▁▁▁▃▅▇●▇▅▄▃▁   #256 of 469     21%      8
math    ▁▁▃▅█▆●█▄▂▁▁   #177 of 431     36%      7
coding · #256 of 469 · 21% of its score Beats [Qwen3-VL 32B Instruct](qwen3-vl-32b-instruct.md), [Solar Open 100B (102B-A12B)](solar-open-100b.md), [GLM-4.7-Flash](glm-4-7-flash.md) here, though they rank higher overall. Trails [DeepSeek LLM 67B Chat](deepseek-llm-67b-chat.md), [QwQ-32B-Preview](qwq-32b-preview.md), [InternLM2.5 20B Chat](internlm2-5-20b-chat.md) here, though they rank lower overall. 
test                                result  vs the others in its run  difficulty        counts  why
LiveCodeBench (version unspecifie…   57.2%  ▄····▄▄··▄●▄ beats 4/6    medium · avg 42%    5.4%  at its level
WebApp1K-Duo-React pass@1 (onekq-…    8.7%  ●··▁·▁▁▁▆█▇█ beats 0/45   hard · avg 29%      4.2%  at its level
WebApp1K-React pass@1 (onekq-ai)     32.3%  ▃▁▁●▁▄▅▃▄▇▁▄ beats 7/31   medium · avg 40%    3.1%  sharp · at its level · noisy
CodeReviewQA: change localisation…   64.6%  █▂▁▁▂▂▂▃▅▅▅● beats 44/47  hard · avg 33%      2.8%  sharp · at its level · noisy
CodeReviewQA: automated code refi…   30.6%  ▁▁▁▁▁▁▂▂▃●▅▄ beats 21/47  hard · avg 32%      2.0%  at its level
CodeReviewQA: solution identifica…   76.4%  ▃··▁▂▂▃▂█▃●▂ beats 39/47  medium · avg 51%    1.9%  sharp · at its level · noisy
LiveCodeBench v2                     71.2%  ▄··█·▄···▄▄● beats 6/7    medium · avg 59%    1.8%  at its level · small run
CodeReviewQA: change type recogni…   73.2%  ▁▁▁······▁▃● beats 24/47  easy · avg 67%      0.1%  blunt · at its level · noisy
math · #177 of 431 · 36% of its score Beats [MiniMax-M2.1](minimax-m2-1.md), [Mistral Medium 3.5 (128B)](mistral-medium-3-5.md), [GPT-4.1](gpt-4-1.md) here, though they rank higher overall. Trails [OpenThinker3 7B](openthinker3-7b.md), [Olmo 3 7B Think](olmo-3-7b-think.md), [OpenThinker2 32B](openthinker2-32b.md) here, though they rank lower overall. 
test                                result  vs the others in its run   difficulty            counts  why
AIME 2024 (maker)                    72.6%  ▃·▃·▃▃···●·▃ beats 6/7     medium · avg 56%         13%  sharp · at its level · several runs
MATH-500 (maker)                     94.3%  ▄·▄▄·▄▄●···▄ beats 5/7     very easy · avg 86%     7.9%  several runs
AIME 2025                            50.0%  ▁·▁▁▁▂▅●█▇▅▁ beats 36/133  medium · avg 37%        7.1%  sharp · at its level
OTIS mock AIME 2024–2025             55.6%  ▃▃▄▃▃▃●▇█▃▂▅ beats 83/197  very hard · avg 7.2%    4.1%  sharp · noisy
OMEGA (math generalisation)          38.9%  ▅···▃▃···●▃█ beats 4/8     hard · avg 24%          1.7%  at its level
THINK-Bench final-answer accuracy…   75.9%  ▄█·●····▄▄·█ beats 3/7     easy · avg 76%          1.2%  typical
MATH (subset unspecified)            92.6%  ▅····▃▃··●▃█ beats 4/8     easy · avg 73%          1.0%  noisy
other tests (7) 
test                                result  vs the others in its run   difficulty            counts  why
GPQA Diamond                         64.1%  ▁▁▁▁▂▄▆●▅▇█▅ beats 93/211  hard · avg 39%           34%  sharp · at its level · several runs
BRIDGE clinical text, zero-shot a…   39.8%  ▁▁▁▁▁▂▂▃▄▆●▃ beats 69/97   hard · avg 31%          4.3%  blunt · at its level
SurgiQ surgical MCQ (zero-shot, l…   60.4%  ▂···▂▂·▃▂▃●█ beats 12/17   medium · avg 53%        1.7%  at its level
EuroEval HellaSwag (English), tes…   79.5%  ▃·▁▁▂▁▂▅█●▄▁ beats 70/100  medium · avg 67%        1.2%  sharp · at its level · noisy
Chess Puzzles (Epoch)                 1.0%  ●▁▁▁▁▂▃▃▃▂▁▁ beats 27/137  very hard · avg 5.8%    0.7%  sharp · too hard for it · noisy
EuroEval Life in the UK (English …   74.5%  ▁▁▁▁▁▁▁▂●█▄▃ beats 48/146  easy · avg 82%          0.6%  noisy
Open PT-LLM Leaderboard average (…   26.2%  ▃▁▁▁▁▁▁▁▂█▅▁ beats 0/276   medium · avg 72%           —  failed run: Failed run: several task sc…
## How the score was worked out Each result implies a score on its own, after adjusting for how hard that test is and who ran it. The model's score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its *counts* column above; here it is added up by who ran the tests. [Method](../method.md) 
source                                    results  share of the score
open-thoughts/OpenThinker2-32B model ca…        4   27% ▇▇▇
open-thoughts/OpenThinker-32B model card        4   17% ▇▇
Epoch AI                                        3   16% ▇▇
deepseek-ai/DeepSeek-R1 model card (sub…        4   15% ▇▇
WebApp1K models leaderboard                     2    7% ▇
CodeReviewQA: The Code Review Comprehen…        4    7% ▇
BRIDGE Medical Leaderboard                      1    4% ▇
allenai/Olmo-3.1-32B-Think model card           2    3% ▇
EuroEval leaderboards                           2    2% ▇
SurgiQ: A Large-Scale Multi-Domain Benc…        1    2% ▇
THINK-Bench: Evaluating Thinking Effici…        1    1% ▇
prior: pull toward the middle                        0% ▇
weighted average                                   = 55.7
Licences: open-thoughts/OpenThinker2-32B model card ([Apache-2.0](https://www.apache.org/licenses/LICENSE-2.0)) · open-thoughts/OpenThinker-32B model card ([Apache-2.0](https://www.apache.org/licenses/LICENSE-2.0)) · Epoch AI ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · deepseek-ai/DeepSeek-R1 model card ([mit](../../licenses/deepseek-ai--DeepSeek-R1.txt)) · WebApp1K models leaderboard ([MIT](../../licenses/webapp1k-leaderboard-MIT.txt)) · CodeReviewQA: The Code Review Comprehension Assessment for Large Language Models ([CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/)) · BRIDGE Medical Leaderboard ([MIT](../../licenses/bridge-medical-leaderboard-MIT.txt)) · allenai/Olmo-3.1-32B-Think model card ([Apache-2.0](../../licenses/allenai--Olmo-3.1-32B-Think.txt)) · EuroEval leaderboards ([MIT](https://github.com/EuroEval/leaderboards/blob/1a77f1d2f8ba1ba05ab73287fb090d76bf41b2e0/LICENSE)) · SurgiQ: A Large-Scale Multi-Domain Benchmark for Evaluating Surgical Understanding in Large Language Models ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · THINK-Bench: Evaluating Thinking Efficiency and Chain-of-Thought Quality of Large Reasoning Models ([CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/))
Not counted (1) 
source                  test                            score  why
Open Portuguese LLM L…  Open PT-LLM Leaderboard avera…  26.2%  failed run: Failed run: several task scores fall far below the trivial baselines (OAB Exams 5.06 on 4-option questions, PT Hate Speech 3.73 and tweetSentBR 3.42 macro-F1) while ASSIN2 RTE is 89.84, i.e. most answers in those tasks were not parsed (reasoning output); the 26.15 average (random baseline 32.64) is not a measurement of ability.
## How to run it ### Through an API api (Cloudflare) · $1.59/M blended · $0.497 in / $4.881 out per 1M tokens · [developers.cloudflare.com](https://developers.cloudflare.com/workers-ai/platform/pricing/) [price source](https://github.com/cloudflare/cloudflare-docs/blob/b182511e39518f4b3304edd7bb3e0b0d3017b035/src/content/docs/workers-ai/platform/pricing.mdx): Workers AI published marginal USD per 1M tokens after free allocation. Workers plan fees excluded; billed in Neurons. Published USD rates are rounded. Product: Workers AI. API model: @cf/deepseek-ai/deepseek-r1-distill-qwen-32b. Settings: plan=Workers Paid, free allocation=excluded, billing=published marginal USD rates; service bills Neurons. Currency: USD. · [context source](https://github.com/anomalyco/models.dev/blob/a668908cfe7073faa0363699486fc41a3ff52412/models/deepseek/deepseek-r1-distill-qwen-32b.toml) ### llama.cpp (recommended) 1. Install llama.cpp. Homebrew and winget installs name the commands llama-cli and llama-server instead. [Other ways to install](https://github.com/ggml-org/llama.cpp/blob/master/docs/install.md) ```sh # macOS / Linux curl -LsSf https://llama.app/install.sh | sh # Windows irm https://llama.app/install.ps1 | iex ``` 2. Run DeepSeek R1 Distill Qwen 32B (Q4_K_M, needs ~19.9 GB, from [huggingface.co/lmstudio-community](https://huggingface.co/lmstudio-community/DeepSeek-R1-Distill-Qwen-32B-GGUF/tree/2c8db776f8037c44c2af1fa197699a0d0c6c4b7a)) ```sh llama cli -hf lmstudio-community/DeepSeek-R1-Distill-Qwen-32B-GGUF:Q4_K_M ``` 3. Or start a local server and open the address it prints (http://localhost:9931) ```sh llama serve -hf lmstudio-community/DeepSeek-R1-Distill-Qwen-32B-GGUF:Q4_K_M ``` Other versions: Q8_0 (~34.8 GB) `llama cli -hf lmstudio-community/DeepSeek-R1-Distill-Qwen-32B-GGUF:Q8_0` ### LM Studio 1. Install LM Studio. Open the app once before using the lms command. [Download it (macOS / Windows / Linux)](https://lmstudio.ai/download) 2. Run DeepSeek R1 Distill Qwen 32B (Q4_K_M, needs ~19.9 GB, from [huggingface.co/lmstudio-community](https://huggingface.co/lmstudio-community/DeepSeek-R1-Distill-Qwen-32B-GGUF/tree/2c8db776f8037c44c2af1fa197699a0d0c6c4b7a)) ```sh lms get https://huggingface.co/lmstudio-community/DeepSeek-R1-Distill-Qwen-32B-GGUF@Q4_K_M -y ``` 3. Then run lms ls to see its name, and lms chat followed by that name Other versions: Q8_0 (~34.8 GB) `lms get https://huggingface.co/lmstudio-community/DeepSeek-R1-Distill-Qwen-32B-GGUF@Q8_0 -y` · 4bit (~18.4 GB, Mac only) `lms get https://huggingface.co/mlx-community/DeepSeek-R1-Distill-Qwen-32B-4bit -y` · 8bit (~34.8 GB, Mac only) `lms get https://huggingface.co/mlx-community/DeepSeek-R1-Distill-Qwen-32B-MLX-8Bit -y` ### Ollama 1. Install Ollama. [Download it (macOS / Windows)](https://ollama.com/download) ```sh # Linux curl -fsSL https://ollama.com/install.sh | sh ``` 2. Run DeepSeek R1 Distill Qwen 32B (Q4_K_M, needs ~20.18 GB, from [ollama.com](https://ollama.com/library/deepseek-r1/tags)) ```sh ollama run deepseek-r1:32b ``` Other versions: Q8_0 (~39.86 GB) `ollama run deepseek-r1:32b-qwen-distill-q8_0` ### MLX (Mac) 1. Install MLX (Mac). Apple silicon Macs only, text only. ```sh # with uv uv tool install mlx-lm # or pip install mlx-lm ``` 2. Run DeepSeek R1 Distill Qwen 32B (4bit, needs ~18.4 GB, Mac only, from [huggingface.co/mlx-community](https://huggingface.co/mlx-community/DeepSeek-R1-Distill-Qwen-32B-4bit/tree/4e0d3848a0ad8f9fb54638891e4928f04fcca978)) ```sh mlx_lm.chat --model mlx-community/DeepSeek-R1-Distill-Qwen-32B-4bit ``` 3. Or start an OpenAI-compatible server at http://localhost:8080/v1 ```sh mlx_lm.server --model mlx-community/DeepSeek-R1-Distill-Qwen-32B-4bit ``` Other versions: 8bit (~34.8 GB, Mac only) `mlx_lm.chat --model mlx-community/DeepSeek-R1-Distill-Qwen-32B-MLX-8Bit` Model repository and documentation: [huggingface.co/deepseek-ai](https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-32B) ## Published speed measurements Reference results for the hardware, engine and settings below. These are not verified speeds of the downloads above or predictions for your machine. 
hardware             precision       speed  engine and settings
Apple M2 Pro, 32 GB  q4_k_m     7.29 tok/s  ollama (version not recorded) · l…
Each speed links to its source, which has the full settings. ## Quantization evidence How much of the full-precision result a quantized version kept, measured within one study. Rows from the same study share a table and aren't independent. They don't describe every download of that precision. 
precision                       test                                full → quantized    kept
AWQ 4-bit (AutoAWQ; vLLM)       AIME 2024 accuracy (mean of 3 pas…     0.644 → 0.678  105.3%
AWQ 4-bit (AutoAWQ; vLLM)       Average of AIME 2024, FOLIO and T…     0.822 → 0.831  101.1%
FP8 W8A8 dynamic (FP8 weights…  MATH-500 pass@1                        0.951 → 0.953  100.2%
FP8 W8A8 dynamic (FP8 weights…  GSM8K 5-shot strict-match (lm-eva…     0.827 → 0.826   99.9%
FP8 W8A8 dynamic (FP8 weights…  OpenLLM v2 average of 6 tasks (IF…     0.369 → 0.369   99.8%
FP8 W8A8 dynamic (FP8 weights…  MMLU 5-shot acc (lm-evaluation-ha…      0.811 → 0.81   99.8%
FP8 W8A8 dynamic (FP8 weights…  OpenLLM v1 average of 6 tasks (AR…     0.745 → 0.744   99.8%
FP8 W8A8 dynamic (FP8 weights…  MMLU-Pro 5-shot (OpenLLM v2 confi…     0.498 → 0.494   99.2%
FP8 W8A8 dynamic (FP8 weights…  Reasoning average of 3 tasks (AIM…     0.763 → 0.755   99.0%
FP8 W8A8 dynamic (FP8 weights…  AIME 2024 pass@1                       0.698 → 0.685   98.2%
FP8 W8A8 dynamic (FP8 weights…  GPQA Diamond pass@1                     0.64 → 0.629   98.2%
GPTQ 3-bit (vLLM)               Average of AIME 2024, FOLIO and T…     0.822 → 0.725   88.2%
GPTQ 3-bit (vLLM)               AIME 2024 accuracy (mean of 3 pas…     0.644 → 0.444   68.9%
GPTQ 4-bit (vLLM)               AIME 2024 accuracy (mean of 3 pas…     0.644 → 0.689  107.0%
GPTQ 4-bit (vLLM)               Average of AIME 2024, FOLIO and T…      0.822 → 0.83  101.0%
INT4 W4A16 (weights only, sym…  GSM8K 5-shot strict-match (lm-eva…     0.827 → 0.844  102.0%
INT4 W4A16 (weights only, sym…  MATH-500 pass@1                         0.951 → 0.95   99.9%
INT4 W4A16 (weights only, sym…  OpenLLM v1 average of 6 tasks (AR…     0.745 → 0.743   99.7%
INT4 W4A16 (weights only, sym…  MMLU 5-shot acc (lm-evaluation-ha…     0.811 → 0.809   99.7%
INT4 W4A16 (weights only, sym…  GPQA Diamond pass@1                     0.64 → 0.638   99.6%
INT4 W4A16 (weights only, sym…  Reasoning average of 3 tasks (AIM…     0.763 → 0.759   99.4%
INT4 W4A16 (weights only, sym…  AIME 2024 pass@1                       0.698 → 0.688   98.7%
INT4 W4A16 (weights only, sym…  MMLU-Pro 5-shot (OpenLLM v2 confi…     0.498 → 0.484   97.2%
INT8 W8A8 (weights symmetric …  GSM8K 5-shot strict-match (lm-eva…     0.827 → 0.839  101.4%
INT8 W8A8 (weights symmetric …  GPQA Diamond pass@1                     0.64 → 0.647  101.1%
INT8 W8A8 (weights symmetric …  OpenLLM v2 average of 6 tasks (IF…      0.369 → 0.37  100.2%
INT8 W8A8 (weights symmetric …  OpenLLM v1 average of 6 tasks (AR…     0.745 → 0.745  100.0%
INT8 W8A8 (weights symmetric …  MATH-500 pass@1                         0.951 → 0.95   99.9%
INT8 W8A8 (weights symmetric …  MMLU 5-shot acc (lm-evaluation-ha…     0.811 → 0.809   99.8%
INT8 W8A8 (weights symmetric …  Reasoning average of 3 tasks (AIM…      0.763 → 0.76   99.6%
INT8 W8A8 (weights symmetric …  MMLU-Pro 5-shot (OpenLLM v2 confi…     0.498 → 0.491   98.6%
INT8 W8A8 (weights symmetric …  IFEval 0-shot instruction-level s…     0.429 → 0.419   97.8%
INT8 W8A8 (weights symmetric …  AIME 2024 pass@1                       0.698 → 0.682   97.7%
Each kept value links to its source. [interactive version](https://whatmodel.sh/model/deepseek-r1-distill-qwen-32b) · [JSON](https://whatmodel.sh/api?format=json&all=1)